Torch: a tool for building topic hierarchies from growing text collections

• Ricardo M. Marcacini USP
• Solange O. Rezende USP

Resumo

Topic hierarchies are very useful for managing, searching and browsing large repositories of text documents. Several studies have investigated the use of hierarchical clustering methods to support the construction of topic hierarchies in an unsupervised way. In this paper we present the Torch - Topic Hierarchies software tool that implements the main features of the method IHTC - Incremental Hierar- chical Term Cluster, aiming to build topic hierarchies from growing text collections. A simple experiment is described in order to illustrate the use of the Torch to support the construction of digital libraries.